Purpose
This study aims to analyse the potential confounding and moderator role of job satisfaction on the effect of working hours on self-perceived health and to analyse the effect of transitions between working hours and job satisfaction.

Design/methodology/approach
Using longitudinal data for the Catalan economy in 2005–2009, first, it runs a linear probability random effects model, with self-perceived health as the dependent variable, on one-year lagged job satisfaction, working hours and its interaction. Second, it estimated an ordered logit model to test the effect of transitions to working hours and different levels of job satisfaction on self-perceived health.

Findings
Short working hours ≤ 20 h/w predict good self-perceived health for women. Long working hours 41–47 h/w predict poor self-perceived health among men and women but not for very long hours ≥ 48 h/w. Interaction effects between working 41–47 h/w and job satisfaction levels were found for men and women. Improvements in job satisfaction for health are reduced when working long hours. For employees, a decrease in job satisfaction may suggest a health risk except if hours also reduce.

Social implications
Workplace practices aimed at gaining flexibility in working hours may be offset, in terms of health outcomes, by lower job satisfaction. Flexible working hours from the employees' side should be favoured to face reductions in job satisfaction.

Originality/value
The novelty of this paper is that highlights differential effect of job satisfaction in the relation between working hours and health status.